Parent-in-the-loop: is it time for AI-supported parental shared decision-making in the neonatal intensive care unit?

Parental decision-making in neonatal intensive care units (NICUs) is often emotionally fraught, with complex multilayered uncertainties that place immense individualistic pressure on parents and clinicians. This paper critically investigates the potential for Applied-Artificial Intelligence (AI) technologies to mitigate this burden by supporting shared decision-making in the NICU. Applied-AI’s predictive capabilities for neonatal prognostic outcomes appear to show promise, but extending this application to the parental decision-making sphere is not straightforward and the technology is at an early largely theoretical stage. We examine ethical, epistemic and relational concerns, identifying risks in: introducing an additional layer of epistemic uncertainty; devaluing parental lived experience and moral agency; exacerbating existing power asymmetries; introducing and expanding existing or hidden biases; implicitly shaping values concerning the infant’s future and negatively affecting the relational dynamics of shared decision-making in the NICU. Awareness of the ethical complexities inherent to the integration of AI tools in this context suggests caution is needed, particularly in circumstances where these tools may give the impression that there is only one ‘obvious’ decision to take. We highlight the value of careful reflection and the importance of developing professional phronesis, that is clinicians learning how to critically and judiciously introduce AI tools in ways that can support wise and ethically sensitive forms of shared decision-making. AI application in neonatal care must preserve and protect the child’s best interests, respect parental autonomy and ensure transparent, human-centred decision-making in the NICU.